CODEWITHSUNDEEP
pythonpandasmultiple-conditions
Laurent Reynaud
Laurent Reynaud

Reputation: 23

Python pandas get previous values row

I created this dataframe with python pandas:

import numpy as np
import pandas as pd

my_df = pd.DataFrame(
    {'ColumnA':['Value A', '', 'Value B', '', '', 'Value C',''],
     'ColumnB':['', '', '', '', '', '', '']})

The dataframe is presented below:

    ColumnA | ColumnB
0   Value A | 
1           | 
2   Value B | 
3           | 
4           |  
5   Value C | 
6           |

To complete column B, I put this condition:

conditions = [
    my_df['ColumnA'] == '',
    my_df['ColumnA'] != '']
result = [my_df['ColumnA'].shift(1),
          my_df['ColumnA']]
my_df['ColumnB'] = np.select(conditions, result)

I have this as a result:

        ColumnA | ColumnB
0     Value A   | Value A
1               | Value A
2     Value B   | Value B
3               | Value B
4               | 
5     Value C   | Value C
6               | Value C

Now I want all the cells in columnB to be filled in like this:

        ColumnA | ColumnB
0       Value A | Value A
1               | Value A
2       Value B | Value B
3               | Value B
4               | Value B
5       Value C | Value C
6               | Value C

Thank you for your suggestions!

Upvotes: 0

Views: 38

Answers (1)

BeRT2me
BeRT2me

Reputation: 13242

Using real NaN values instead of empty strings makes many things easier...

import pandas as pd
import numpy as np

data = {'ColumnA':['Value A', '', 'Value B', '', '', 'Value C',''],
        'ColumnB':['', '', '', '', '', '', '']}
df = pd.DataFrame(data)

# Fix your null values:
df = df.replace('', np.nan)

# You appear to want ColumnB to be ColumnA if it were forward-filled.
df['ColumnB'] = df['ColumnA'].ffill()

print(df)

Output:

   ColumnA  ColumnB
0  Value A  Value A
1      NaN  Value A
2  Value B  Value B
3      NaN  Value B
4      NaN  Value B
5  Value C  Value C
6      NaN  Value C

Upvotes: 1

Related Questions